For someone who is accustomed to going almost daily to the animal room with a shopping list of mice needed for experiments, it is hard to imagine that there had been a time when no inbred strains existed. It is even more difficult to envision how the world would have looked had inbred strains never been invented. Although the resourceful human mind would probably have found ways to eventually arrive at our present level of knowledge even without the strains, such an alternate course, no doubt, would have been much more arduous. Among the disciplines hardest hit by the lack of inbred strains would have been immunogenetics and, in particular, transplantation genetics. We would have learned about the existence of histocompatibility genes without the strains, but how we would have accumulated all the current knowledge about the individual genes, particularly the genes in the major histocompatibility complex, we cannot even guess. We are, therefore, happy to join in this tribute to the pioneers who initiated inbreeding of mouse strains. They have put into our hands a tool which for contemporary biology is as indispensable as the scintillation counter, the pH meter, or the Sephadex column. They have enabled the construction of new scientific disciplines probing into new territories of modern biology. For all this we are extremely grateful and it is our pleasure to share with them today some of the excitement that they have helped to generate.
The work that we are going to describe started with inbred-strain analysis but it soon led us to the original source of all such strains -- the wild mice. It concerns the H-2 complex, the major histocompatibility complex of the mouse ( Figure 1). When we began the work some ten years ago, Peter Gorer, Bernard Amos, Gustavo Hoecker, and George Snell had already established that among the inbred strains there were some ten alleles, or haplotypes as they are now referred to, at this complex locus (for review see reference 1). This degree of genetic variability was surprising since only two or three alleles were known at all other loci identified at that time. Was this variability of the H-2 complex a true polymorphism or was it merely the result of chance fixation? And if the H-2 complex was more polymorphic than other loci, what was then the meaning of the polymorphism, and how did the polymorphism relate to the function of the H-2 loci? What was the function of the H-2 complex anyway? These questions were provocative enough to stimulate our interest in the study of H-2 variability.
The first step in this study was to trap wild mice in farm buildings, bring them into the laboratory, and type them with antisera defining individual antigens controlled by the H-2 complex. All such antisera available to us at that time were produced by crossimmunization of inbred strains and they defined H-2 antigens of the inbred strains. The typing of the wild mice with the antisera established two important facts, but provided only partial answers to our original questions ( Figure 2). The two facts were, first, that some antisera were broadly reactive with wild mice, while others reacted with such mice hardly at all ( 2), and second, that two mice from a single locality resembled each other in their H-2 antigens more closely than mice from different localities ( 3, 4). The broadly cross-reactive antisera were against H-2 antigens which were also shared by many inbred strains and which we therefore designated as public ( 2). The reactivity pattern of these antisera suggested that the wild mice carried H-2 haplotypes different from those present in the inbred strains, but we could not be completely sure about this conclusion because it is the private(the most restricted antigen) that unambiguously characterizes a haplotype. The antisera tot eh private H-2 antigens reacted hardly at all with the wild mice, and if they did, it often turned out that the reactivity was caused by a previously undetected public antigen. When antibodies to the new antigen were removed by absorption, the antiserum ceased to react with the particular wild mouse while it still reacted with the inbred strain carrying the private antigen. It was a tantalizing finding: all we could get was a hint that wild mice possessed H-2 haplotypes absent in inbred strains, but we could neither prove this variability nor could we determine how many new haplotypes were present among the mice.
The finding of locality-dependent diversification of H-2 haplotypes was interesting and it fit well with observations made earlier by ecologists about the social structure of the mouse populations. Mice apparently live in small family units (demes) which are relatively isolated and migration from one unit to another is probably rare ( 5, 6). The similarity of H-2 haplotypes within a single locality could be, therefore, the result of the genetic isolation of the families.
The ambiguity of wild-mice typing with antisera prepared against inbred H-2 haplotypes convinced us that the way of going about characterizing wild H-2 haplotypes was to produce antisera against antigens controlled by these haplotypes. This may sound like a simple thing to do, but in fact there were problems with the approach. First of all, there was the question of how to get enough material for immunization. A single wild mouse was not going to provide enough tissue for the large number of injections necessary to produce an H-2 antiserum. Then, there was the problem of how to test the antiserum when the donor of the immunizing tissue was no longer available. And finally, we feared that immunization with so genetically heterogeneous a donor as the wild mouse would produce a mixture of antibodies that would take us forever to sort out.
Because of these problems we decided to use a different approach. We resolved to isolate a sample of H-2 haplotypes from wild mice and place them on the well-defined genetic background of an inbred strain. In other words, we wanted to do what George Snell did earlier with so many inbred H-2 haplotypes -- to produce H-2 congenic lines. Because most of Snell's congenic lines were on the background of strain C57BL/10Sn, or B10 for short, we decided to use this strain for the crossing with wild mice, and to produce a series of B10.W congenic lines (where W stands for "wild"). It was then that we got the first taste of what it was like in the pioneering days of inbred and congenic strain production! For the task of producing 30 or 40 B10.W lines looks simple on paper, but in reality one has to overcome all kinds of mishaps to get the mice through the ten or 12 generations of backcrossing required for congenic line production. Now, more than ever, we can appreciate what was involved when the entire program of inbred-strain production was begun!
To produce the B10.W lines, we took advantage of the fact that in the middle of the H-2 complex is a locus coding for a serum protein, the serum serological substance or Ss ( Figure 1, reference 7). Because most wild mice carry the Ssh allele at this locus, while some inbred strains carry the Ssl allele, in the B10.W-production system we would backcross repeatedly to an Ssl Ssl strain (B10.BR) and always select for further mating animals carrying the Ssh allele and thus presumably the wild-derived alleles at the H-2 loci on both sides of Ss ( Figure 3).
We started the project with some 80 B10. W lines; now, some ten years later, when the project is almost completed, we ended up with 35 lines. The others were all lost because of poor breeding performance.
Once the production of the B10.W lines was completed, we took the next step -- to analyze their H-2 haplotypes. To this end we tested the lines with a battery of antisera defining all the known private H-2 antigens. We then immunized inbred strains [usually the (A x B10)F1 hybrids] with B10.W tissues, produced anti-B10.W sera, and tested the antisera against all inbred strains carrying different H-2 haplotypes, and against all other B10.W lines. Finally, we did an absorption analysis with all strains and all antisera that were positive in a direct test. The analysis consisted of absorbing each positive antiserum with each positive strain and testing on all positive strains. Although we tried to avoid the production of antibodies to public H-2 antigens in the anti-B10.W sera by choosing the A strain as one of the parents of the F1 hybrid recipient (the H-2a haplotype of strain A codes for most of the known public antigens), we were not completely successful. The broad cross-reactivity of the anti-B10.W indicated that there were many more public antigens than were previously defined and that wild mice carried a number of public antigens absent in inbred strains. Because we were interested only in antigens that characterize the new public antigens. On the contrary, we tried to remove the anti-public antibodies from our antisera and to restrict the reactivity pattern of the antisera as much as we could.
The analysis of all the B10.W strains has not been completed yet. Thus far, however, we have been able to identify 20 new H-2 haplotypes and 25 new H-2 antigens ( Table 1), all of which are absent in the inbred strains ( 8, 9, 10, 11, 12, 13). We designated the new haplotypes H-2w1 through H-2w20 and the new antigens H-2.101 through H-2.125. In addition, we identified three new H-2 haplotypes in strains carrying alleles at the t locus ( 11) which is positioned in the same chromosome as H-2 and which affects embryonic and sperm development ( 14). These haplotypes we designated according to the t allele they carry, that is, H-2twl, H-2t12, and H-2tw5.
Table 2 lists all of the H-2 haplotypes -- inbred or wild -- identified thus far. The total number is 109. Some of these haplotypes, however, share alleles at both H-2K and H-2D (they differ at other loci of the H-2 complex not discussed in this communication). If we exclude these shared haplotypes, we still end up with an impressive number of 68 different haplotypes. Thus, among inbred strains and wild mice, 68 different combinations of alleles at the H-2K and H-2D loci have already been identified.
Some of the haplotypes in Table 2 are recombinants, that is, they possess alleles at H-2K or H-2D loci also present in other haplotypes but in different combinations. Much of the variability of the H-2 complex is thus generated by combinatorial means with alleles at the two H-2 loci existing in many different combinations. Theoretically, the number of different h-2K-H-2D combinations equals the number of alleles at the H-2K locus multiplied by the number of alleles at the H-2D locus. The alleles at the two loci are listed in Table 3: there are 37 of them at the H-2K locus and 32 at H-2D. The total number of possible H-2K-H-2D combinations generated by the known alleles is, therefore, 1,184.
As mentioned earlier, the H-2 complex contains several loci in addition to H-2K and H-2D. The total number of loci composing the complex is not yet known, but ten loci have already been identified ( Figure 1). Although most other H-2 loci are probably far less polymorphic in comparison with H-2K and H-2D, they do contain multiple alleles. And so, even if one assumes that there are just a few alleles at other H-2 loci, the number of possible H-2 haplotypes reaches almost astronomical proportions.
The development of the B10.W lines was meant to be just a springboard for our analysis of wild mouse populations. Because the private inbred H-2 antigens were so poorly represented among wild mice, we hoped to identify private antigens of wild mice by means of our B10.W lines. As Table 1 shows, we have succeeded in doing so. Once we produced antisera defining new private or semi-private H-2 antigens, we went back to wild mouse populations with this considerably enriched battery of typing reagents to determine the frequencies of individual H-2 antigens. This time the battery contained antisera against 36 H-2 antigens ( Table 4), of which about one-half were characterized as private and the rest as semi-private (the latter being shared by a limited number of strains available to us).
So far we have typed 70 wild mice, but the work continues and data on a much large panel should be available soon. The mice were trapped at the different localities in Texas, most of them in various farm buildings; only a few were trapped in the fields (One should bear in mind that the wild mice from which our B10.W lines were developed were trapped in Michigan and thus represented a geographically very distant population with regard to the Texas mice).
To calculate phenotypic frequencies of the 36 H-2 antigens defined by our battery of antisera, we also used the panel of B10.W lines and took each line as a representative of one wild mouse. The total number of wild mice typed is therefore 90 (70 trapped mice and 20 B10.W lines).
The calculations of phenotypic frequencies were carried out by dividing the number of mice typed as positive for a given antigen by the number of mice tested ( 90). The frequencies calculated in this manner are subject to several qualifications.
In the first place, we ignored the fact the groups of mice came from the same localities and that the presence in a given locality of several mice positive for a given antigen probably reflected close blood-relationship among such mice. The justification for ignoring the origin of the mice was our hope that in a large sample covering wide geographical area the local effects on antigen frequencies would be minimized. However, the sample described here might not have been large enough and local effects might have biased our antigen-frequency calculations.
In the second place, we did not take into consideration the H-2 zygosity of the wild mice. A mouse typed as negative for a given antigen does not pose any problem in this respect, but a positive mouse could be either homo- or heterozygous for the antigen. We are in a process of establishing the frequency of H-2 heterozygosity among wild mice and once we have the figure, we should be able to use it as a correction factor in our frequency estimates. For the calculations in Table 4, such a factor has not yet been introduced and the data should be therefore viewed with this fact in mind. It is also important to realize that the data in Table 4 represent phenotypic and not gene frequencies.
The third qualification concerns the pooling of data on wild mice and B10.W line typing. Each B10.W line is, in fact, not equivalent to one wild mouse because it carries only one of the two H-2 haplotypes of the original wild mouse from which it was derived. Consequently, by typing wild mice we test for two haplotypes (provided that the mouse is an H-2 heterozygote), while by typing B10.W lines we test for only one haplotype. However, since in most instances the antigen frequencies calculated separately for wild mice and B10.W lines do not differ significantly, the pooling of the data is probably justified.
The last qualification concerns the specificity of the reagents in the typing battery. Each antiserum, before it was included in the battery, was extensively analyzed against panels of inbred strains and B10.W lines and absorbed so that it reacted only against a given antigen, but we could never be sure that it did not contain additional antibodies. Such antibodies might have escaped our detection when the antiserum was tested against the strain panels, but they might have emerged when the serum was tested on wild mice. In one or two instances we know that this actually did happen. For example, antiserum detecting antigen H-2.16 was monospecific in our panel tests, that is, it reacted only with strains carrying H-2.16 and no other mice. But when the antiserum was tested against our sample of 70 wild mice, it reacted with an unexpectedly high number of them. Furthermore, some mice positive with this antiserum were also positive for several other private H-2 antigens and this observations, of course, aroused our suspicion that the anti-H-2.16 might not be monospecific after all. And indeed, when the antiserum was absorbed by some of the positive wild mice, the reactivity against these mice was removed but the serum still reacted with H-2.16 positive strains. Apparently, despite all efforts to make it monospecific, the antiserum still contained at least two antibodies, one against antigen H-2.16 and another against a public antigen which is present in H-2.16-positive strains and some wild mice, but absent in all other strains in our typing panels. Once we had learned about the second antibody, we made sure that for future testing an anti-H-2.16 serum was used from which the antibody was absorbed out, but we of course could not tell how many of the previously typed positive wild mice actually carried H-2.16 and how many reacted with the antiserum because of the presence of the previously unidentified antibody to the public antigen. We solved this problem by ignoring the results obtained with the antiserum before its complexity was discovered.
This example, however, raises the question of how many of our other typing antisera are complex in a similar way. To answer this question we would have to do absorption analysis with each positive mouse, but such analysis with so many reagents and so many mice to test is not humanly possible. We do, however, spot check all the antisera by absorbing them with wild mice and testing them for their monospecificity, and we check, of course, all antisera for additional antibodies when our suspicion is aroused. But despite all these precautions, one must first consider the antigen-frequency data in Table 4 as possibly overestimated. The frequencies of at least some of the antigens might be actually lower than our preliminary data indicate.
Among the private H-2 antigens in Table 4 the following are detected by antisera which we have no reason to suspect as containing more than one antibody: H-2.2, 4, 15, 17, 18, 19, 20, 21, 23, 26, 30, 31, 32, 33, 106, 108, 109, 110, 111, 112, 113, and 114. The other antigens are either semi-private ones or antigens defined by antisera that might contain additional antibodies. Several points can be made about these preliminary data.
The most frequent among the "clean" private antigens is H-2.31 which was found in six wild mice and three B10.W strains (phenotypic frequency of some 10%). At the opposite end of this scale are antigens H-2.15 and H-2.18 which we have not found in any of the typed wild mice and which are also absent in all the B10.W lines. The antigens are carried by inbred strains and one might speculate that these strains were derived from wild mice unrelated to the population of wild mice that we are currently typing.
The majority of the clean private antigens occur in the Texas wild mice and in the B10.W lines with frequencies ranging from 1 to 5%. If we assume -- as our data suggest -- that aside from H-2.31, there are not very many frequently occurring H-2 antigens and that the average H-2 antigen frequency is 2%, then we may estimate that the population we are presently testing may contain some 50 antigens controlled by the H-2K locus and another 50 antigens controlled by the H-2D locus. Barring linkage disequilibrium, these numbers would then lead to the estimate of 502 or 2,500 haplotypes (or more precisely H-2K-H-2D allelic combinations).
It is interesting to compare these estimates with the frequencies of antigens controlled by the human homologue of the H-2 complex -- the HLA system. There are now some 15 antigens known to be controlled by the HLA-A locus and another 20 antigens controlled by the HLA-B locus ( 15). About half of these HLA antigens occur in frequencies similar to the phenotypic frequency of H-2.31 (i.e., 10 to 20%), while the other half of HLA antigens occurs in frequencies of 5% or less, that is, frequencies similar to those of the majority of H-2 antigens. In other words, in the HLA system there seem to be more antigens occurring in high frequencies than in the H-2 complex. This difference between the H-2 and HLA systems could mean one of two things: either some of the HLA antisera defining the more common antigens are not truly monospecific and the difference between H-2 and HLA is the consequence of differences in methodologies, or the H-2 complex is more polymorphic than the HLA system.
Some 43% of the wild mice in our sample failed to react with any of the antisera in our battery. These "blanks" indicate the presence in the wild mice of H-2 antigens against which we have not yet made antisera. Thus there are probably more antigens than we have been able to identify so far, and search for these antigens is in progress.
One can also expect that mice from other parts of the world might have still different antigens. A comparison of North American, European, and Asiatic mice might provide some interesting results in this respect.
In addition to providing, for the first time, information about H-2 antigen frequencies and about the extent of the H-2 polymorphisms, our wild mice studies are proving to be useful in other areas of H-2 immunogenetics, primarily as a source of new H-2 variants. Several of our B10.W lines turned out to be H-2 recombinants, carrying haplotypes that are new combinations of H-2K and H-2D alleles present in other strains. Such an enrichment of the H-2 recombinant collection should be helpful for all kinds of H-2 mapping studies.
Other B10.W strains seem to carry alleles which are similar to but not identical with alleles carried by other strains. The difference between these alleles is very likely the consequence of recent mutations. Comparison -- particularly a biochemical one -- of such natural variants should provide much needed information as to how the variability of the H-2 loci is generated.
And so in our studies we have traveled the full circle: we began with inbred strains, extended from there to wild mice, and now with our B10.W lines we have come back to inbred strains again. And we have been able to travel the circle in such a relatively short time only because we has a solid foundations in the lines produced by the fathers of mouse inbreeding.
We thank Mr. Wayne Bundy for breeding most of the strains mentioned in this communication, and Ms. Jeanne Lively for secretarial help.
1. Klein, J. (1975). Biology of the Mouse Histocompatibility-2 Complex Springer-Verlag, New York.
See also MGI.
2. Klein, J. (1971). Nature 229: 635.
See also PubMed.
3. Klein, J. (1970). Science 168: 1362.
See also PubMed.
4. Klein, J., and Bailey, D. (1971). Genetics 68: 287.
See also PubMed.
5. Crowcroft, P.L. (1955). J. Mammal. 36: 299.
6. Reimer, J., and Petras, M. (1967). J. Mammal. 48: 88.
See also PubMed.
7. Shreffler, D. (1964). Genetics 49: 973.
See also PubMed.
8. Klein, J. (1972). Transplantation 13: 291.
See also PubMed.
9. Klein, J. (1975). Immunogenetics 2: 297.
10. Hauptfeld, M., and Klein, J. (1976). Immunogenetics 3: 603.
11. Hammerberg, C., Klein, J., Artzt, K., and Bennett, D. (1976). Transplantation 21: 199.
See also PubMed.
12. Zaleska-Rutczynska, Z., and Klein, J. (1977). J. Immunol. 119: 1903.
See also PubMed.
13. Klein, J., and Zaleska-Rutczynska, Z. (1977). J. Immunol. 119: 1912.
See also PubMed.
14. Klein, J., and Hammerberg, C. (1977). Immunol. Rev. 33: 70.
See also MGI.
15. Snell, G., Dausset, J., and Natherson, S.G. (1977). Histocompatibility. Academic Press, New York.